Log In Sign Up

Demystify Self-Attention in Vision Transformers from a Semantic Perspective: Analysis and Application

by   Leijie Wu, et al.

Self-attention mechanisms, especially multi-head self-attention (MSA), have achieved great success in many fields such as computer vision and natural language processing. However, many existing vision transformer (ViT) works simply inherent transformer designs from NLP to adapt vision tasks, while ignoring the fundamental difference between “how MSA works in image and language settings”. Language naturally contains highly semantic structures that are directly interpretable by humans. Its basic unit (word) is discrete without redundant information, which readily supports interpretable studies on MSA mechanisms of language transformer. In contrast, visual data exhibits a fundamentally different structure: Its basic unit (pixel) is a natural low-level representation with significant redundancies in the neighbourhood, which poses obvious challenges to the interpretability of MSA mechanism in ViT. In this paper, we introduce a typical image processing technique, i.e., scale-invariant feature transforms (SIFTs), which maps low-level representations into mid-level spaces, and annotates extensive discrete keypoints with semantically rich information. Next, we construct a weighted patch interrelation analysis based on SIFT keypoints to capture the attention patterns hidden in patches with different semantic concentrations Interestingly, we find this quantitative analysis is not only an effective complement to the interpretability of MSA mechanisms in ViT, but can also be applied to 1) spurious correlation discovery and “prompting” during model inference, 2) and guided model pre-training acceleration. Experimental results on both applications show significant advantages over baselines, demonstrating the efficacy of our method.


page 3

page 4

page 6

page 7


A Survey on Visual Transformer

Transformer is a type of deep neural network mainly based on self-attent...

Learning Multiscale Transformer Models for Sequence Generation

Multiscale feature hierarchies have been witnessed the success in the co...

SCRAM: Spatially Coherent Randomized Attention Maps

Attention mechanisms and non-local mean operations in general are key in...

IA-RED^2: Interpretability-Aware Redundancy Reduction for Vision Transformers

The self-attention-based model, transformer, is recently becoming the le...

Human Interpretable AI: Enhancing Tsetlin Machine Stochasticity with Drop Clause

In this article, we introduce a novel variant of the Tsetlin machine (TM...

Rethinking Positional Encoding in Language Pre-training

How to explicitly encode positional information into neural networks is ...

eX-ViT: A Novel eXplainable Vision Transformer for Weakly Supervised Semantic Segmentation

Recently vision transformer models have become prominent models for a ra...